Immunologic Recovery in Children after Alternative Donor Allogeneic Transplantation for Hematologic Malignancies: Comparison of Recipients of Partially T Cell–Depleted Peripheral Blood Stem Cells and Umbilical Cord Blood

Abstract

Impaired immunologic recovery (IR) after hematopoietic stem cell transplantation (HSCT) is associated with increased risk for infections and relapse. Stem cell source and graft manipulation influence the kinetics of IR. Partial T cell depletion of peripheral blood stem cell (PBSC) grafts is a novel alternative method of graft manipulation for children. We compared IR in children undergoing HSCT for hematologic malignancies receiving either T cell–depleted (TCD)-PBSCs (n = 55) or umbilical cord blood (UCB) (n = 21) over a 7-year period at a single institution. PBSC grafts underwent ex vivo negative selection for CD3⁺ cells using the Clin-iMACS system with partial T cell add-back. Recovery of CD4⁺ T cells was significantly delayed in TCD-PBSC recipients compared with UCB recipients, owing to impaired CD4⁺/CD45RA⁺ (naïve) T cell lymphopoiesis. Recovery of total CD3⁺ cells and CD3⁺/CD8⁺ cells was similar in the 2 groups. The TCD-PBSC recipients had a marked deficit in CD19⁺and, to a lesser extent, IgA/IgM, owing to the need for B cell depletion of these grafts to attenuate the risk of lymphoproliferative disease after TCD HSCT. There were no significant between-group differences in response to mitogen stimulation, time to independence from intravenous immunoglobulin supplementation, or incidence of viral reactivation. Transplantation outcomes of relapse, transplantation-related mortality, event-free survival, and overall survival were similar in the 2 groups. Efforts to enhance IR after partial TCD-PBSC transplantation, such as selective αβ T cell depletion, hold promise for further improvement of this transplantation approach.

INTRODUCTION

Allogeneic hematopoietic stem cell transplantation (HSCT) is the sole curative option for many children with high-risk hematologic malignancies, but only approximately 30% of these children will have a matched related donor (MRD). To broaden access to this treatment modality, the use of alternative donors is increasing, including bone marrow (BM) or mobilized peripheral blood stem cells (PBSCs) from matched unrelated or partially matched related donors, as well as unrelated umbilical cord blood (UCB). Immunologic recovery (IR) after alternative donor allogeneic HSCT in children is complex and dynamic, influenced by various patient and transplantation-related factors, including age of recipient and donor, indication for transplantation, conditioning regimen, donor type, stem cell source, graft manipulation, infection, and graft-versus-host disease (GVHD) chemoprophylaxis, type, and treatment [1,2]. Impaired IR increases the risk of serious infection [3] and relapse [4], and is associated with decreased survival [5].

Donor type and stem cell source affect IR [2]. The type and quantity of passenger lymphocytes infused as a component of the graft differ by stem cell source and degree of manipulation. Passenger lymphocytes provide initial lymphoid immunity after undergoing thymus-independent homeostatic peripheral expansion [6]. BM and PBSC grafts contain predominately memory Tcells, whereas UCB grafts have a higher proportion of naive T cells with differing immunobiology [7]. Graft manipulation with either positive or negative cell selection influences this early thymus-independent lymphoid recovery by altering the cellular composition of the graft [8–10].

In the second wave of IR that occurs several months after transplantation, phenotypically naïve T cells that have undergone maturation in the thymus emerge [11]. This thymic-dependent process is heavily influenced by various clinical factors known to affect thymic function, including age [2], conditioning regimen (particularly irradiation), and presence of GVHD [12]. Because these factors are inextricably linked to donor type and cell source, the kinetics of IR continue to be influenced by graft characteristics even late after transplantation.

At our institution, all types of alternative donors are considered, and the ultimate decision depends on HLA matching, urgency, patient size, and other factors (eg, cyto-megalovirus [CMV] status). In an effort to maintain the benefits of mobilized PBSCs—including rapid neutrophil and platelet engraftment [13]—while mitigating the increased risk of cGVHD owing to greater numbers of T cells in the graft [14], we have used partial T cell depletion. Our current method involves ex vivo negative selection for CD3⁺ cells using the CliniMACS system (Miltenyi Biotec, Bergish-Gladbach, Germany) with partial T cell add-back. To evaluate IR in recipients of alternative donor grafts, we compared IR after HSCT using this graft type with a concurrent cohort of UCB recipients—the other major alternative donor source at our hospital. Institutional preference for PBSCs with CD3⁺ depletion has resulted in relatively few unrelated donor BM transplantations, so this group was not included because meaningful statistical comparison was precluded by small numbers. In addition, we focused our analysis on alternative donor HSCT, given that IR after MRD HSCT has been extensively characterized and is the preferred approach when available. Comparison of IR between alternative graft types may inform decisions regarding donor selection when an MRD is unavailable.

MATERIALS AND METHODS

Patients and Transplantation Regimens

We retrospectively reviewed the charts of 76 consecutive children undergoing first allogeneic HSCT for a hematologic malignancy at our hospital between March 2005 and December 2011. During this period, 55 children received TCD-PBSC grafts that had been CD3⁺-depleted using the CliniMACS system (clinicaltrials.gov identifier NCT00579124), and 21 patients received UCB grafts. The protocol was approved by our hospital’s Institutional Review Board, and written informed consent was obtained from all patients and/or parents, as appropriate.

All patients with an acute hematologic malignancy were in morphological remission (complete remission [CR]) at the time of transplantation. Myeloablative conditioning was provided either with cyclophosphamide 60 mg/kg/day for 2 days and total body irradiation (TBI) 200 cGy twice daily for 3 days (80.3%) or with cyclophosphamide and busulfan 0.8 to 1 mg/kg every 6 hours for 16 doses, adjusted to achieve a target steady-state concentration of 750 to 1100 ng/mL (19.7%), both either with (81.6%) or without thiotepa 5 mg/kg/day for 2 days.

A calcineurin inhibitor was administered as primary GVHD chemoprophylaxis in all patients. UCB recipients also received methylprednisolone 1 mg/kg/day starting on day +7 and tapered starting at day +21, as well as granulocyte-colony stimulating factor (G-CSF; filgrastim) until the absolute neutrophil count exceeded 2000 cells/μL. In the absence of GVHD, immunosuppression was tapered starting on day +100 (or earlier if there was concern for declining chimerism). All patients received standard infectious prophylaxis that was individually tailored to risk. Specifically, for CMV-positive UCB recipients and CMV-positive TCD-PBSC recipients with CMV-negative donors, foscarnet was used until engraftment occurred, at which point it was replaced with valganciclovir, along with IVIG supplementation. Otherwise, CMV prophylaxis was provided with IVIG supplementation alone. Weekly plasma polymerase chain reaction testing for adenovirus, CMV, and Epstein-Barr virus (EBV) was performed on all patients from day +7 up to day +100 and as indicated thereafter; testing for additional viruses was based on clinical indications.

Graft Manipulation

TCD-PBSC grafts were obtained by leukapheresis of peripheral blood mononuclear cells after G-CSF stimulation. CD3⁺ depletion was performed by negative selection using the automated CliniMACS device, as described previously [15,16]. To mitigate the risk of post-transplantation lymphoproliferative disease, all PBSC grafts underwent some form of B cell depletion, either ex vivo or in vivo. Before the availability of beads conjugated to anti-CD19 antibodies, 38 patients received rituximab 375 mg/m²/dose on days −1 and +7; thereafter, grafts were depleted of B cells during ex vivo manipulation (n = 17). The number of CD3⁺ cells added back at the time of the stem cell product infusion was determined by the degree and nature of HLA mismatching and disease status, ranging from 0.2 to 8 × 10⁵ cells/kg of recipient weight (median, 1 × 10⁵ cells/kg). The majority of patients (67%) received between 1 × 10⁵ and 3 × 10⁵ cells/kg. In general, patients with high-risk leukemia, less HLA disparity, and malignancies associated with greater graft-versus-leukemia effects (eg, chronic myelogenous leukemia [CML]) received higher CD3⁺ cell doses. Of the 7 patients who received >3 × 10⁵ cells/kg, 4 were 10/10 HLA matches, and all had CML, juvenile myelomonocytic leukemia, or acute leukemia with evidence of minimal residual disease at the time of transplantation. Recipients of low CD3⁺ cell doses generally had ≥2 HLA antigen mismatches and were in durable CR.

Measurement of Immunologic Recovery

Formal assessments of immunologic recovery were made at 4, 8, 12, and 24 months after HSCT as the standard of care. Monoclonal antibodies to surface antigens were used in flow cytometry analysis to define the following immunophenotypes: CD3⁻/CD16⁺ and/or CD56⁺ (natural killer [NK] cells), CD3⁺ (T cells), CD3⁺/CD4⁺ (CD4⁺ T cells), CD3⁺/CD8⁺ (cytotoxic T cells), CD4⁺/CD45RA⁺ (naïve CD4⁺ T cells), CD4⁺/CD45RO⁺ (memory CD4⁺ T cells), CD19⁺ (B cells), and CD20⁺ (B cells). From these data, the ratios of CD4⁺ cells to CD8⁺ cells (4:8) and that of naïve cells to memory cells (RA:RO) were calculated. Immunoglobulins (IgG, IgA, and IgM) were measured at the same intervals by standard nephelometry. The time to independence from intravenous gamma globulin (IVIG) supplementation was based on the interval between day 0 and the last dose of IVIG, in days. Patients received supplemental IVIG until the IgG level was maintained at >500 mg/dL without support.

The response of peripheral blood mononuclear cells to stimulation with the mitogens phytohemagglutinin (PHA), pokeweed (PWM), and concanavalin A (ConA) was measured at 8, 12, and 24 months post-transplantation. Results are reported as ratios of patient to control counts per minute (CPM_rel); a normal response was considered to be ≥50% of normal (CPM_rel ≥0.5).

Among the subjects alive at 1 year, published age-based ranges in healthy children for number of total lymphocytes, CD3⁺ cells, CD16⁺/CD56⁺ cells, CD3⁺/CD4⁺ cells, CD3⁺/CD8⁺ cells, and CD19⁺ cells were used to define patients who were in at least the 5th percentile as “normal” [17].

Definitions of Transplantation Outcomes

Neutrophil engraftment was defined as the first of 3 consecutive days on which the peripheral blood absolute neutrophil count was ≥500 cell/uL, and platelet engraftment was defined as the first of 7 consecutive days with an unsupported platelet count ≥20,000 cells/uL. Pre-engraftment bacteremia was defined as any positive blood culture obtained between the time of initiation of the conditioning regimen and neutrophil engraftment. Viral reactivations were included only if they were considered clinically significant, as defined by either the need for antiviral therapy or the presence of associated clinical manifestations. Untreated reactivations or infections without clinical disease (eg, human herpesvirus-6 [HHV6] viremia without symptomatic disease) were not counted.

Staging and grading of acute GVHD (aGVHD) and chronic GVHD (cGVHD) was based on Center for International Blood and Marrow Transplant Research guidelines. Patients who developed GVHD after donor lymphocyte infusion (DLI) were considered to have GVHD. The time to cessation of immunosuppressive therapy (IST) was defined as the first time after HSCT or diagnosis of aGVHD or cGVHD at which the patient achieved freedom from immunosuppressive medications for at least 1 month.

Relapse was defined as morphological evidence of recurrent disease in the peripheral blood or BM. Patients with mixed chimerism who responded to DLI were not considered to have relapsed. Transplantation-related mortality (TRM) was defined as all nonrelapse deaths. Event-free survival was calculated based on the following events: relapse, TRM, or DLI.

Statistical Analysis

Baseline demographic and transplant characteristics were compared using Fisher exact tests for categorical data and Wilcoxon rank-sum for continuous variables. The outcome variables of absolute lymphocyte numbers, 4:8 and RA:RO ratios, immunoglobulin levels, and mitogen responses were logarithmically (base 10) transformed to obtain data showing symmetric distribution. All analyses of the primary outcome measures of these variables were performed on these transformed values. A linear mixed effects model was fitted to each outcome variable using Proc Mixed in SAS 9.2 (SAS Institute, Cary, NC). The model included group, time, and group-by-time interaction as fixed effects, along with a random intercept and a random slope for each patient. This approach accounts for potential correlations among repeated measurements, and was used to test differences between TCD-PBSC and UCB recipients and differences between patients with and without cGVHD.

Two multivariate models were constructed to control for potential confounders. The first model incorporated pretransplantation variables with a known association with IR after HSCT: age, sex, disease category (acute lymphoblastic leukemia, acute myelogenous leukemia, other), conditioning regimen (TBI, no TBI), HLA disparity (any mismatch, fully matched), and receipt of antithymocyte globulin. A second model incorporated all of these variables with the addition of cGVHD, the sole post-transplantation variable included in multivariate analysis.

Differences in the proportion of patients reaching age-based normal lymphocyte levels and mitogen response (as defined above) were assessed using the Fisher exact test. The following separate exposures were used: UCB and TCD-PBSC; aGVHD grade II–IV and no aGVHD/grade I; cGVHD and no cGVHD; and use of IST and no IST at 1 year.

The Kaplan-Meier method was used to estimate overall survival (OS) and event-free survival (EFS), and differences between groups were tested using the log-rank statistic. Patients who were event-free were censored at the time of last follow-up. Cumulative incidence curves were generated for relapse and TRM, adjusting for the other outcome as a competing risk. Differences between the groups were tested using Gray’s test [18].

All statistical tests were 2-sided, with a significance level of P < .05. All analyses were performed using Stata 12.1 (StataCorp, College Station, TX) or SAS 9.2.

RESULTS

Demographic and Transplantation Characteristics

The 2 groups were well matched overall for major demographic and transplantation-related variables (Table 1). However, as expected, the median age at transplantation was significantly younger in UCB recipients than in TCD-PBSC recipients (5 years versus 11 years; P < .001). In addition, TCD-PBSC recipients were more likely to receive a TBI-containing preparative regimen (P = .003). Antithymocyte globulin was administered to 8 UCB recipients (38.1%) as part of conditioning (proximal) and 4 TCD-PBSC recipients (7.3%) several weeks before transplantation (distal) in patients who had not received previous chemotherapy (eg, patients with myelodysplastic syndrome or CML) (P = .003).

Table 1.

Demographic and Transplantation Characteristics

Characteristic	All Patients	TCD-PBSC	UCB	P Value
Number	76	55	21
Age, yr, median (IQR)	9 (5–14)	11 (7–16)	5 (1–7)	<.001
Female sex, n (%)	39 (51.3)	28 (50.9)	11 (52.4)	NS
Ethnicity, n (%)				NS
White	44 (57.9)	30 (54.6)	14 (66.7)
Black	10 (13.2)	10 (18.2)	0
Other	22 (29.0)	15 (27.3)	7 (33.3)
Disease, n (%)				NS
Acute lymphoblastic leukemia	30 (39.5)	22 (40.0)	8 (38.1)
Acute myelogenous leukemia	19 (25.0)	13 (23.6)	6 (28.6)
CML	7 (9.2)	7 (12.7)	0
Myelodysplastic syndrome	15 (19.7)	10 (18.2)	5 (23.8)
JMML	3 (4.0)	2 (3.6)	1 (4.8)
Bilineage	1 (1.3)	1 (1.8)	0
Other	1 (1.3)	0	1 (4.8)
Remission status, n (%)				NS
CR1	27 (50.9)	22 (57.9)	5 (33.3)
CR2+	26 (49.1)	16 (42.1)	10 (66.7)
Conditioning, n (%)				.003
Cyclophosphamide/TBI ± thiotepa	61 (80.3)	49 (89.1)	12 (57.1)
Busulfan/cyclophosphamide ± thiotepa	15 (19.7)	6 (10.9)	9 (42.9)

JMML indicates juvenile myelomonocytic leukemia; NS, not significant.

P values are based on Wilcoxon rank-sum or Fisher exact tests.

Graft Characteristics

Graft characteristics are summarized in Table 2. The overall proportion of HLA-mismatched donor–recipient pairs was similar in the 2 groups, but the TCD-PBSC recipients had more HLA-DRB1 mismatches. Total nucleated and CD34⁺ cell doses were within expected ranges based on stem cell source. The median CD3⁺ cell dose was more than 100-fold lower in TCD-PBSC grafts compared with UCB grafts.

Table 2.

Graft Characteristics

Characteristic	TCD-PBSC	UCB
Donor type, n (%)
Matched sibling donor	0	1 (4.8)
Unrelated donor	49 (89.1)	20 (95.2)
Mismatched related donor	6 (10.9)	0
HLA disparity, n (%)
Any mismatch	39 (70.9)	17 (81.0)
≥2 loci	24 (43.6)	10 (47.6)
DR mismatch	6 (10.9)	0
Graft composition, cells/kg, median (IQR)
Total nucleated cells	2.52 × 108 (3.43–7.08 × 108)	0.51 × 108 (0.24–0.72 × 108)
CD34 cells	6.5 × 106 (4.92–8.74 × 106)	0.41 × 106 (0.19–0.5 × 106)
CD3 cells	1 × 105 (1–2.67 × 105)	1.2 × 107 (0.6–2.3 × 107)

Engraftment and Innate Immunity

Neutrophil recovery occurred significantly earlier in the TCD-PBSC recipients compared with the UCB recipients (median, 15 days versus 19 days; P < .001), as did platelet engraftment (median, 18 days versus 35 days; P < .001). All patients experienced myeloid engraftment, but 3 TCD-PBSC recipients (5.5%) and 2 UCB recipients (9.5%) did not achieve platelet engraftment owing to early TRM. NK cell numbers were significantly higher in UCB recipients at all time points, but both groups exhibited robust normal early levels that gradually trended down over time after HSCT (Figure 1A).

Figure 1.

Levels of NK cells (A), total lymphocytes (B), and T cells (C) as a function of time after HSCT, by graft type. All lymphocyte numbers have been logarithmically transformed and are presented as mean with 95% CI error bars. P values based on linear mixed-effect modeling: ^*P < .001; ^#P < .01; ^{^}P < .05.

Adaptive Immunity

UCB recipients had significantly higher total lymphocyte numbers compared with TCD-PBSC recipients up to 2 years post-HSCT (Figure 1B). Although more UCB recipients than TCD-PBSC recipients attained age-specific normal total lymphocyte levels by 1 year post-HSCT, the differnece was not statistically significant (78.6% versus 60.5%; P = .33) (Table 3).

Table 3.

Proportion of Patients Reaching Normal Levels by 1 Year by Graft Type

	TCD-PBSC (n = 34)	UCB (n = 14)	P Value
Cell type
Absolute lymphocyte count	60.5	78.6	.33
CD3+	50	78.6	.11
CD3+/CD4+	67.7	78.6	.51
CD3+/CD8+	70.6	64.3	.74
CD19+	64.7	92.9	.073
Mitogen
PHA	44.0	54.6	.56
PWM	52	45.5	.72
ConA	60.0	63.6	.84

Data are percentage of patients alive at 1 year with normal lymphocyte levels or mitogen response. For lymphocyte subsets, normal is defined as ≥5th percentile of age-based normal levels. For mitogen response, normal is defined as ≥50% of the control response. All P values are based on the Fisher exact test.

T cells

Absolute numbers of T cells were comparable in the 2 groups at all time points (Figure 1C); however, by 1 year post-HSCT, only 50.0% of TCD-PBSC recipients had achieved normal T cell levels—the lowest percentage of any cell type—compared with 78.6% of UCB recipients (P = .11) (Table 3). Within the T cell compartment, absolute levels of cytotoxic T cells (Figure 2B) and the proportion reaching normal levels by 1 year were nearly identical in the 2 groups (Table 3). In contrast, CD4⁺ T cell numbers were lower in TCD-PBSC recipients up to 2 years post-HSCT (Figure 2A). Given the relative paucity of CD4⁺ T cells in TCD-PBSC recipients, the 4:8 ratio was generally lower than that seen in UCB recipients. In both groups, this ratio trended downward (to more normal levels) as cytotoxic T cells gradually recovered (Figure 2E).

Figure 2.

Levels of CD4⁺ T cells (A), cytotoxic T cells (B), naïve CD4⁺ T cells (C), and memory CD4⁺ T cells (D), and 4:8 ratio and RA:RO ratio as a function of time after HSCT (E), by graft type. All lymphocyte numbers have been logarithmically transformed and are presented as means with 95% CI error bars. 4:8 and RA:RO are untransformed and presented as mean with SEM error bars. P values based on linear mixed-effect modeling: ^*P < .001; ^#P < .01; ^{^}P < .05.

Within the CD4⁺ T cell compartment, naïve T cell recovery was impaired in TCD-PBSC subjects—persisting up to 2 years post-HSCT—whereas memory T cell recovery was comparable in the 2 groups (Figure 2C and D). The ratio of naïve to memory CD4⁺ T cells (RA:RO ratio) reflects these trends. Although both groups showed a gradual increase in the RA:RO ratio as naïve T cells emerged (presumably after thymic maturation), this increase was earlier and more robust in the UCB recipients, resulting in a significantly higher ratio at 8 and 12 months post-HSCT (Figure 2F).

Humoral immunity

The most pronounced difference between the 2 groups was in B cells, which were profoundly impaired in the TCD-PBSC recipients, in terms of both absolute numbers (Figure 3A) and the proportion of patients achieving normal levels (Table 3). This deficit was seen in both pre-Ig and post-Ig class-switched memory B cell subsets (CD19⁺/CD27⁺/IgD⁺ and CD19⁺/CD27⁺/IgM⁺, respectively). Of note, in the TCD-PBSC recipients, memory B cells (CD27⁺) did not demonstrate evidence of substantial reconstitution until 1 year post-HSCT, whereas in UCB recipients, levels began to increase as early as 8 months post-HSCT and demonstrated a consistent linear trend (data not shown).

Figure 3.

Levels of B cells (A), IgM (B), and IgA (C) as a function of time after transplant, by graft type. All lymphocyte numbers have been logarithmically transformed and are presented as means with 95% CI error bars. P values based on linear mixed-effect modeling: ^*P < .001; ^#P < .01; ^{^}P < .05.

IgM levels were lower in TCD-PBSC recipients than in UCB recipients up to 2 years post-HSCT (Figure 3B). IgA levels converged earlier, with relatively comparable levels as soon as 8 months, although the between-group difference remained significant until 1 year (Figure 3C). The average time to independence from IVIG supplementation was 157 days in the TCD-PBSC recipients and 178 days in the UCB recipients (P = .73).

Response to mitogens

In contrast to the sometimes marked differences in absolute numbers of various lymphocyte subsets, the response to mitogen stimulation was remarkably similar in the 2 groups. As shown in Figure 4, there was no statistically significant difference in response to any mitogen between the groups at 8, 12, or 24 months post-HSCT, regardless of mitogen. Although some variation, generally functional recovery (≥50% of control) was achieved in about half of patients in both groups by 1 year (Table 3). Notably, TCD-PBSC recipients seemed to perform comparably to UCB recipients even in response to mitogens requiring competent B cell function (PWM).

Figure 4.

Responses to mitogen stimulation based on graft time by time after HSCT, by graft type. (A) PHA. (B) PWM. (C) ConA. Data are presented as mean CPM_rel with SEM error bars. No significant between-group differences were detected.

GVHD and IST

aGVHD

Rates of any aGVHD were similar in the 2 groups (UCB, 66.7%; TCD-PBSC, 76.4%; P = .537). However, UCB recipients tended to have higher incidences of grade II–IV and grade III–IV aGVHD (61.1% and 33.3%, respectively) compared with TCD-PBSC recipients (47.3% and 18.2%, respectively), although neither of these differences was statistically significant.

Because the presence of grade II–IV aGVHD did not affect the numbers of any of the lymphocyte subpopulations measured, with the exception of B cells (data not shown), and did not affect the proportion of patients reaching normal levels by 1 year (Table 4), aGVHD was not included in the multivariate analysis.

Table 4.

Proportion of Patients Achieving Normal Lymphocyte Levels and Mitogen Response at 1 Year, by GVHD Status

	aGVHD			cGVHD			IST
	Grade 0–I (n = 22)	Grade II–IV (n = 25)	P Value	None (n = 22)	Limited/Extensive (n = 27)	P Value	No (n = 32)	Yes (n = 16)	P Value
Cell type
Absolute lymphocyte count	61.5	68.0	.63	68.2	59.3	.52	71.9	50.0	.14
CD3+	50.0	64.0	.33	57.1	56.0	.94	63.3	42.9	.20
CD3+/CD4+	77.3	64.0	.32	85.7	56.0	.029	83.3	42.9	.006
CD3+/CD8+	68.2	68.0	.99	66.7	68	.92	66.7	71.4	.75
CD19+	77.3	68.0	.48	90.5	56.0	.019	83.3	50.0	.021
Mitogen
PHA	50.0	42.1	.64	62.5	33.3	.089	50	44.4	.78
PWM	56.3	47.4	.60	68.8	33.3	.039	58.3	44.4	.48
ConA	56.3	63.2	.68	68.8	50	.27	66.7	55.6	.56

Data are percentage of patients alive at 1 year with known GVHD status with normal lymphocytes levels or mitogen response. For lymphocyte subsets, normal is defined as ≥5th percentile of age-based normal levels. For mitogen response, normal is defined as ≥50% of the control response. P values are based on the Fisher exact test.

cGVHD and immunosuppression

The incidence of extensive cGVHD in evaluable patients was similar in the 2 groups (TCD-PBSC, 15.2%; UCB, 21.4%; P = .69), reflected in the nearly identical proportions of patients in the 2 groups (approximately 30%) who remained alive and on IST at 1 year post-HSCT. The median time to cessation of IST was 195.5 days (IQR, 127.5 to 400.5 days) in TCD-PBSC recipients and 159 days (IQR, 129 to 456 days) in UCB recipients (P = .96). Eight TCD-PBSC recipients required DLI owing to declining chimerism or frank relapse. All 4 patients who were salvaged and subsequently attained durable CR experienced cGVHD (extensive in 2 patients).

Patients with cGVHD and those receiving IST at 1 year post-HSCT were statistically significantly less likely than those without cGVHD and off IST to achieve normal levels of CD4⁺ T cells and B cells by 1 year (Table 4). A comparison of lymphocyte numbers in patients with cGVHD and those without cGVHD showed lower levels of naïve CD4⁺ T cells at 12 and 24 months in affected patients and lower levels of B cells at all times in affected patients up to 2 years post-HSCT (Figure 5B and C). Absolute numbers of other lymphocytes did not differ significantly by cGVHD status, even when the entire CD4⁺ T cell pool was examined (Figure 5A).

Figure 5.

Levels of CD4⁺ T cells (A), naïve CD4⁺ T cells (B), and B cells (C) as a function of time after HSCTon, by cGVHD status. All lymphocyte numbers have been logarithmically transformed, and are represented as means with 95% CI error bars. P values based on linear mixed-effect modeling: ^*P < .001; ^#P < .01; ^{^}P < .05.

The response to stimulation with PWM at 1 year post-HSCT was severely reduced in patients with cGVHD compared with those without cGVHD. There was a trend toward the same effect after PHA stimulation, although this did not reach statistical significance (Table 4). This discrepancy in the effect of cGVHD on mitogen response is likely attributable to the fact that PWM response reflects both B cell and T cell function, whereas PHA is more T cell–specific.

Multivariate Analysis

After adjustment using both models, the between-group differences in absolute numbers of naïve CD4⁺ T cells and B cells, IgM, IgA, and RA:RO persisted at all previously significant time points. Differences in NK cells, CD4⁺ T cells, and 4:8 ratio were abrogated in both multivariate models. This suggests that the differences in CD4⁺ T cell and NK cell recovery between UCB and TCD-PBSC recipients are mediated by factors beyond graft source or manipulation.

Infection and Transplantation Outcomes

Infection

The UCB recipients had a higher rate of pre-engraftment bacteremia than the TCD-PBSC recipients, although the difference was not statistically significant (28.6% versus 16.4%; P = .33). Although most viral reactivations occurred early (median, day +36.5; IQR, day +19 to day +71) and thus reflect early IR, we compared the incidences of CMV, EBV, HHV6, and adenovirus in the 2 groups as a proxy for quality of overall IR. Rates of recipient CMV seropositivity and incidence of CMV reactivation were comparable in the TCD-PBSC and UCB recipients (18.2% versus 14.3%), as were the incidences of HHV6 (23.6% versus 23.8%) and EBV (3.6% versus 4.8%). All patients with EBV reactivation were successfully preemptively treated with rituximab before clinically evident lymphoproliferative disease developed; these patients were included in the analysis of B cell reconstitution. Adenovirus infection was more common in the UCB recipients (19.1% versus 5.5%; P =.087). When all viral infections or reactivations were examined in aggregate, the incidence of any viral infection was nearly identical in the TCD-PBSC and UCB recipients (43.6% versus 47.6%).

Relapse, TRM, and survival

There were no between-group differences in the competing risk-adjusted cumulative incidence of either relapse or TRM, or in the Kaplan-Meier survivor probabilities of OS and EFS. The 2-year OS was 65.2% (95% confidence interval [CI], 50.3% to 76.6%) in the TCD-PBSC recipients and 60.6% (95% CI, 36.1% to 78.2%) in the UCB recipients, and the 2-year EFS was 48.5% (95% CI, 34.2% to 61.4%) in the TCD-PBSC recipients and 54.6% (95% CI, 30.3% to 73.5%) in the UCB recipients.

DISCUSSION

Quality IR after HSCT is critical for protection against both infection and relapse, and is associated with improved survival. Stem cell source and graft manipulation are known to influence the kinetics of IR [19–23]. The advantages of PBSCs as a stem cell source are achieved at the expense of a higher incidence of cGVHD [13,14]. Conversely, UCB grafts generally are associated with a low incidence of GVHD even in the face of greater HLA disparity, but with slower engraftment and higher risk of graft failure [24,25]. The present study compared the kinetics of IR in pediatric recipients of a relatively novel graft type (partial TCD-PBSC) and recipients of a more established alternative donor source (UCB). These 2 graft types represent the predominant alternative donor sources at our institution and are used in patients with high-risk malignancies when an HLA-matched sibling donor is unavailable.

Strengths of this study include the relatively large sample sizes, long duration of follow-up, and homogenous transplantation regimens in the 2 study groups, who underwent HSCT for similar indications and received comparable conditioning and supportive care. This permitted a focused evaluation of the effects of stem cell source and ex vivo manipulation on IR.

T cell recovery was generally impaired after TCD-PBSC HSCT, but a close examination of critical T cell subsets elucidates key disparities between the 2 study groups. TCD-PBSC recipients demonstrated a pronounced naïve CD4⁺ T cell lymphopenia that accounted for much of the between-group disparity in CD4⁺ T cell IR. Moreover, despite 2-log fewer passenger T cells in TCD-PBSC grafts, the number of memory T cells was similar in TCD-PBSC recipients and UCB recipients at all time points, likely related to the relative paucity of passenger memory T cells in UCB grafts. Because much of the early T cell expansion after HSCT is driven by homeostatic peripheral expansion of passenger lymphocytes infused with the graft, the immunobiology of the graft and the process of ex vivo graft manipulation has important implications for the kinetics of early IR [8–10]. In PBSC (and BM) grafts, these passenger lymphocytes characteristically express a memory phenotype; however, UCB grafts contain a relatively higher percentage of naïve T cells [7]. Our results are consistent with these principles.

In a second wave of T cell lymphopoiesis—typically occurring 8 to 12 months after HSCT—thymic-dependent maturation occurs, producing naïve T cells with diverse TCR repertoires. Clinical factors known to affect thymic function, such as age [2,26], GVHD [12,27], and the conditioning regimen, strongly influence this process. Both groups experienced an increase in the naïve T cell pool between 4 and 8 months, resulting in a prominent increase in the RA:RO ratio at this time; however, this phenomenon was more pronounced in the UCB recipients, suggesting superior thymic-mediated naïve T cell recovery. It is possible that this difference is confounded by the fact that compared with TCD-PBSC recipients, UCB recipients are younger and less likely to have received TBI-based conditioning, both of which influence thymic function. However, differences in naïve CD4⁺ T cells persisted in the multivariate analysis which adjusted for patient age and TBI exposure, suggesting the graft biology as the overriding driving force.

Exposure of donor lymphohematopoietic cells to G-CSF during mobilization is known to influence cytokine release and immunologic function, possibly contributing to the CD4⁺ cell deficit seen in TCD-PBSC recipients. Preclinical data indicate that the balance of CD4⁺ subsets—including Th1, Th17, and regulatory T cells—can vary greatly and potentially influence the development of GVHD in animal models [28]. Because TCD is nonselective, future studies evaluating the proportion of each subset within the CD4⁺ compartment and the post-HSCT homeostatic expansion of these subsets will be of significant clinical interest. Moreover, our findings are limited by our assumption that naïve CD4⁺ T cells represent recent thymic emigrants; however, more specific methods exist for defining this population of thymically matured T cells, including TCR rearrangement excision circles, further immunophenotyping (eg, CCR7⁺, CD62L⁺), and spectratyping of the TCR repertoire diversity [2].

In contrast to these between-group differences in the CD4⁺ T cell pool, recovery of cytotoxic T cells was equivalent in the 2 groups. However, because UCB recipients are known have impaired cytotoxic T cell recovery compared with other unrelated donors [29,30], this still likely represents a major deficiency in IR after TCD-PBSC HSCT.

The pattern of B cell recovery demonstrates a profound and prolonged relative B cell lymphopenia in TCD-PBSC recipients compared with UCB recipients across all B cell subsets. The UCB recipients demonstrated rapid and robust quantitative B cell recovery, with more than 90% achieving normal levels by 1 year, in agreement with previous reports [24,31]. Compounding this disparity is the need to deplete the TCD-PBSC grafts of B cells to mitigate the risk of EBV-related lymphoproliferative disease. Early rituximab exposure after HSCT has been shown to profoundly affect humoral immunity [32]. Somewhat reassuringly, differences in IgM and IgA levels were less pronounced, and the time to independence from IVIG supplementation was similar in the 2 groups.

Notably, despite these disparate patterns of lymphocyte and immunoglobulin recovery, there were no major between-group differences in the incidence of early viral reactivation or proliferative response of PBMC to mitogen stimulation, suggesting comparable functional immunity. Both of these measures serve as surrogates for immune function and reflect both humoral and cellular immunity [33]. Response to mitogen stimulation was notably similar even with PWM, which requires B cell competency, and with ConA, a mitogen associated with high sensitivity in identifying T cell immunodeficiencies [33]. A lag between quantitative lymphocyte recovery and actual functional recovery after transplantation has been well reported, but a more pronounced lag after UCB HSCT owing to the “immaturity” of passenger graft immune cells could possibly explain our findings. Alternatively, if the passenger lymphocytes in a TCD-PBSC graft contain a high proportion of central memory cells, then the initial homeostatic peripheral expansion may be sufficient to confer early protection against viruses, as would passive infusion of high numbers of effector memory cells, which would not be expected to expand significantly [34]. Moreover, NK cells contribute to early antiviral (and antitumor) protection as well as to engraftment [35], and these cells recovered early in both groups. Early and robust expansion of NK cells is also seen in autologous HSCT and may help explain the low incidence of viral reactivation in these recipients as well. Our data on the incidence of viral reactivation are limited by the fact that most viral reactivations occurred early (median, day +36.5), before our first assessment of IR.

GVHD exerts deleterious effects on lymphocytes by direct lymphocyte toxicity [36], impairment of thymic function [12,27], and its treatment with IST. We did not detect a major effect of aGVHD on immunologic recovery, despite universal treatment of grade III–IV aGVHD with 2- or 3-drug regimen that included systemic steroids. We speculate that this lack of association can be attributed to the relatively late initial assessment (at 4 months) of IR in our study. Moreover, much of the aGVHD seen in the present study was steroid-responsive and did not necessarily require additional IST or long-term steroid use. On the other hand, cGVHD significantly impaired CD4⁺ T cell and B cell recovery. Within the CD4⁺ T cell compartment, naïve T cells seem to be preferentially suppressed by cGVHD. This is likely attributable in part to direct thymic damage by alloreactive cellular mediators (with resulting impairment of thymic-dependent naïve cell reconstitution), supported by the fact that differences do not emerge until after 8 months post-HSCT when thymic-dependent lymphopoiesis becomes a major contributor to IR. In addition, this effect likely is partially mediated by pharmacologic treatment of cGVHD, given that patients who remained on IST at 1 year post-HSCT had even more pronounced CD4⁺ T cell deficits. The deleterious effect of cGVHD on B cells as a result of both direct toxicity and reduced presence of support cells has been described as well [37].

Despite apparent differences in certain aspects of IR, the major clinical outcomes of relapse—TRM, EFS, and OS—were similar in our 2 groups of HSCT recipients. This suggests that the process of TCD does not negate the graft-versus-leukemia effects of PBSC HSCT and that the impaired IR seen in these graft recipients does not substantively increase the risk of infection-related TRM compared that in with UCB recipients. The use of this graft manipulation method holds promise for effective application of HSCT in children with high-risk hematologic malignancies who lack an HLA-matched related donor and who do not have an 8/8 or 10/10 high-resolution unrelated donor match; almost three-quarters of our TCD-PBSC recipients were mismatched at ≥ 1 HLA locus, and many were mismatched at multiple loci. However, ongoing efforts to enhance IR after TCD-PBSC HSCT, such as selective depletion of αβ⁺ T cells [38], are important to further improve outcomes using this transplantation approach.